Numerical and Experimental Characterisation of Convective Transport in Solid Oxide Fuel Cells

Thumbnail Image
Resch, Emmanuel
Solid Oxide Fuel Cell , Pressure Effects , SOFC , Convection , Permeability , Model , CFD , Klinkenberg , Knudsen , Electrode
In this work, numerical and experimental methods are used to characterise the effects of convective transport in an anode-supported tubular solid oxide fuel cell (SOFC). To that end, a computational fluid dynamics (CFD) model is developed to compare a full transport model to one that assumes convection is negligible. Between these two approaches, the variations of mass, temperature, and electrochemical performance are compared. Preliminary findings show that convection serves to reduce the penetration of hydrogen into the anode, and becomes more important as the thickness of the anode increases. The importance of the permeability of SOFC electrodes on the characterization of convection is also investigated. Experiments performed on Ni-YSZ anodes reveal that permeability is a function of the cell operating conditions, and increases with increasing Knudsen number. An empirical Klinkenberg relation is validated and proposed to more accurately represent the permeability of electrodes in a CFD model. This is a departure from an assumption of constant permeability that is often seen in the literature. It is found that a varying permeability has significant effects on pressure variation in the cell, although according to the electrochemical model developed in this work, variation in permeability is only found to have minor effects on the predicted performance. Furthermore, it is revealed that an electrochemical model which makes the simplifying assumption of constant overpotential is in error when predicting current and temperature variation. In this work, this is found to predict an unrealistic spatial variation of the current. It is suggested that this approach be abandoned for the solution of a transport equation for potential which couples the anodic and cathodic currents. This will lead to a more realistic prediction of temperature and performance.
External DOI